gistfile1.txt

class MyClass
{
    function __construct(Collaborator %collaborator)
    {
    }
}

// vs.

class MyClass
{
    function doSomething(Collaborator $collaborator)
    {
    }
}

Yeah I get the idea, but what is the reason behind it?
Saving time by writing less constructor logic?

Or does MyClass have various methods and this one method requires an extra dependency that isn't injected using the construct?

MyClass is composed by SomeHttpAction, which creates an implementation of Collaborator depending on some input parameter. SomeHttpAction and MyClass are created at the startup, while Collaborator is created after analyzing the request: that's why I find it handy to inject it as a parameter; from the point of view of MyClass nothing really changes.